← Anthropic
Matt Wolfe
AI News: Fable Banned, New Open-Source Leader, Midjourney Shocker
⏱ 36 min video · 3 min read19 Jun 2026Worth watching
TL;DR
Matt Wolfe covers the biggest AI news of the week, including the US government forcing Anthropic to shut down Claude Opus 4 (called 'Fable 5' in the transcript) over a security jailbreak dispute, a powerful new open-source model from ZAI (GLM 5.2) rivaling frontier models at a fraction of the cost, and Midjourney's surprising pivot into medical body-scanning spas.
Key points
1
The US government forced Anthropic to shut down its most advanced models (Claude Opus 4 / Fable 5 and Mythos 5) for all global users after Anthropic dismissed a reported jailbreak as minor, with Amazon CEO Andy Jassy reportedly being the whistleblower despite Amazon being a major Anthropic investor.
2
ZAI released GLM 5.2, a 753-billion-parameter open-weight MIT-licensed model with a 1M token context window that ranks #2 on the web dev arena leaderboard behind Claude Fable 5, at roughly $1.40 input / $4.40 output per million tokens versus Claude Opus at $5-25.
3
Midjourney launched Midjourney Medical, an ultrasound-based body scanner using ~9,000 transducers in a water tank, with plans to open a spa in San Francisco in 2027 where customers can get full-body scans alongside hot tubs and saunas.
4
OpenAI rolled out a 'Record and Replay' feature in Codex that records a user performing a task on-screen, learns the steps, and can repeat that task automatically in the future.
5
Perplexity introduced self-improving memory ('Brain') for its Computer agent, which reviews past work overnight, learns from corrections, and continuously optimizes its performance without user input.
Key takeaways
→
GLM 5.2 is currently free to use at chat.z.ai and delivers near-frontier coding performance, making it worth testing as a cost-effective alternative to Claude or GPT-5 for agentic and coding tasks.
→
Investors in AI foundation labs should treat the Anthropic shutdown as a precedent-setting risk signal: even a company valued near $1 trillion can be forced to pull its flagship product overnight by government action.
→
The Anthropic situation illustrates a credibility trap: publicly advocating for government AI regulation (as Dario Amodei did in his essay) creates political and regulatory exposure when the company later resists a specific government request.
Notable quotes
“Dario did spend months saying that this is really, really scary. It could wreak havoc in the wrong hands and that this stuff absolutely needs to be regulated. And now the loudest advocate for AI regulation is frustrated that it's getting regulated.”
“It's frankly bewildering that Anthropic hasn't wanted to comply with safety requests that it previously said were its highest priority.”
“Drink from the fire hose. I'll be overwhelmed so you don't have to.”
Worth watching?
✅
Worth watching the full video?
Watch if you want the live GLM 5.2 coding demo and Midjourney Medical video breakdown - the key facts are all here, but the visual comparisons and hands-on testing add useful context.
Saved you some time? The creator still deserves a like.
Want this for your own channels?
Add the channels you follow. Every new video summarised and in your inbox the moment it drops. From £4/month.
Try it free